Genetica (2011) 139:699–707 DOI 10.1007/s10709-011-9578-9

SI - GOS

Current and historical hybridization with differential introgression among three species of cyprinid fishes (genus )

Richard E. Broughton • Krishna C. Vedala • Tessa M. Crowl • Lauren L. Ritterhouse

Received: 6 August 2010 / Accepted: 18 April 2011 / Published online: 4 May 2011 Ó Springer Science+Business Media B.V. 2011

Abstract Hybridization is common among freshwater Introduction fishes, particular among the . We used two mitochondrial genes and one nuclear gene to characterize Natural zones of hybridization have been called windows hybridization among two species pairs of Cyprinella in on the evolutionary process (Harrison 1990) because they southwestern North America. Genalogical patterns represent cases where speciation is incomplete and thus revealed that C. lutrensis and C. venusta are currently may reveal the genetic and ecological factors that lead to hybridizing in several localities producing apparent F1,F2 reproductive isolation while they are still in progress. and backcross generations, yet there was no evidence for Hybridization appears to be more common among fresh- introgression outside of local hybrid zones. Alternatively, water fishes than in other vertebrate groups and hybrids are mitochondrial haplotypes from C. lutrensis appear to have relatively common in nature (Hubbs 1955; Campton 1987; introgressed into a C. lepida population in the Nueces Scribner et al. 2001). Hybridization may increase diversity River completely replacing the native C. lepida haplotype. at the population level, at least temporarily, through There was no evidence of introgression of nuclear DNA development of hybrid swarms and via reciprocal intro- and there does not appear to be ongoing hybridization. The gression of alleles into the genetic background of parental population of C. lepida from the nearby Frio River exhibits populations. Alternatively, strongly asymmetric introgres- no evidence of hybridization with C. lutrensis. Thus, con- sion may reduce diversity by overwhelming the genomes tact between C. lutrensis and C. venusta results in the of recipient species (potentially leading to extinction), formation of localized hybrid swarms, while contact particularly when there is a corresponding asymmetry in between C. lutrensis and C. lepida has resulted in complete population size (e.g., Echelle and Connor 1989; Dowling mitochondrial introgression in the Nueces River or no and Childs 1992). If hybridizing populations merge into a apparent hybridization in the Frio River. The three differ- single interbreeding unit, taxonomic diversity is reduced ent outcomes of contact between these species illustrate the (generating one species from two). Another possible out- variable nature of interspecific reproductive interactions come is isolation of a hybrid population from its parent and provide an excellent system in which to better under- species yielding an entirely new taxon, increasing taxo- stand the factors influencing hybridization among fresh- nomic diversity (three species from two) (e.g., DeMarais water fishes. et al. 1992). The nature of reproductive barriers appears to be dynamic, as reflected by spatial and temporal variation Keywords Cyprinidae Hybridization Introgression in the degree of reproductive (e.g., Dowling et al. 1997; Aboim et al. 2010). While hybridization appears to be an important component of freshwater fish evolution (Smith R. E. Broughton (&) K. C. Vedala T. M. Crowl 1992; Dowling and Secor 1997; Gerber et al. 2001), our L. L. Ritterhouse understanding of the factors that promote or allow frequent Oklahoma Biological Survey and Department of Zoology, interspecific mating remains limited. University of Oklahoma, 111 E Chesapeake St, Norman, OK 73019, USA Hybridization has been documented in a variety of e-mail: [email protected] freshwater fish taxa, including black basses and sunfishes 123 700 Genetica (2011) 139:699–707

(Centrarchidae; Bolnick and Near 2005), darters (Percidae; (e.g., Lytle 1972; Matthews 1987; Mayden 1989). Phylo- Keck and Near 2010), mollies and guppies (Poeciliidae; genetic studies revealed two distinctive mtDNA haplotypes Turner et al. 1980, Scribner et al. 2001), and and from C. lepida, one found in the upper Nueces River and chubs (Cyprinidae; Dowling et al. 1989, Alves et al. 2001; the other from the Frio and Sabinal Rivers which form a Aboim et al. 2010). Cyprinidae is the worlds largest clade tributary to the Nueces (Richardson and Gold 1995; of freshwater fishes and is also the largest family in North Broughton and Gold 2000). Despite occurring as little as America with 275–280 species (Mayden et al. 1992). 50 km apart, individuals from these two areas were so Interspecific hybridization is common in the diverse North different from each other that Richardson and Gold (1995) American cyprinid fauna and is not restricted to matings suggested that the Nueces population was an undescribed among closely related species. For example, Hubbs (1955) species. Haplotypes from C. lepida were phylogenetically reported extensive hybridization between members of dif- similar to those from different populations of C. lutrensis, ferent cypinrid genera and subgenera. Sixty-eight species rendering the latter paraphyletic (Broughton and Gold pairs found east of the continental divide produce inter- 2000). Although not initially apparent, this pattern is con- generic or inter-subgeneric hybrid offspring (Hubbs 1955). sistent with genetic exchange between C. lepida and More recently, Smith (1992) observed that morphological C. lutrensis (Schonhuth and Mayden 2010; this paper). divergence provides little constraint on the ability of North Photographs of C. lepida from the Frio and Nueces Rivers American cyprinids to interbreed. The extent of hybrid- appear in Fig. 1. ization among more closely related cyprinid congeners has Here, we describe the genealogical history of mito- not been completely documented but anecdotal observa- chondrial and nuclear genes from several southwestern tions suggest it is likely to number in the hundreds of populations of C. lutrensis, C. venusta and C. lepida. species pairs. This suggests that much of the divergence Population samples were drawn mainly from three drain- that leads to ‘‘species-level’’ differences occurs in the ages in close proximity in south Texas where the three absence of reproductive isolation and that intrinsic barriers species occur. The history of hybridization may be revealed to gene exchange may often arise independently of mor- by the extent of introgression among recent hybrid indi- phological differences. viduals and the depth at which haplotype or allelic lineages Cyprinella is a monophyletic group of 30 cyprinid are resolved on the genealogy. If hybridization is recent species that are among the most locally abundant fishes in and/or ongoing individuals of mixed ancestry should pos- streams and rivers throughout eastern North America sess cross-specific gene variants reflecting local variation (Mayden 1989). Cyprinella lutrensis is the most wide- in parental populations. Alternatively, historical hybrid- spread species, occurring throughout the Great Plains ization events may be marked by discordance among loci and arid southwest. Cyprinella venusta is also broadly dis- reflecting differential introgression and sorting among lin- tributed, occurring in most Gulf-coastal drainages from eages that arose deeper in genealogical history. Our results Florida to northern Mexico. The ranges of C. lutrensis and clarify the evolutionary history of C. lepida and charac- C. venusta overlap in Texas, southern Oklahoma, and terize genetic aspects of hybridization between C. lutrensis Louisiana (Mayden 1989). Although not frequently and C. venusta, provide new insights in the dynamic nature encountered at the same locality, individuals of mixed and variable results of hybridization in North American ancestry have been observed at several localities in Texas cyprinid fishes. where they co-occur (Hubbs and Strawn 1956). Experi- mental crosses between C. lutrensis and C. venusta indi- cated that hybrid individuals are fertile, capable of Materials and methods backcrossing with both parental species and producing F2 hybrids (Hubbs and Strawn 1956). Hybridization between Individual fishes were collected by seine and preserved in these species has been observed more recently where 95% ethanol in the field. Table 1 lists sample localities and C. lutrensis has been introduced outside its native range in the species collected as well as the number of individuals Coosa River system of Alabama (Walters et al. 2008; Blum sequenced for nuclear and mitochondrial genes. Rivers and et al. 2010). Figure 1 illustrates phenotypes of each species. sampling localities are illustrated in Fig. 2. Sampling Cyprinella lepida has a very restricted range, occurring focused on C. lepida from the Nueces River and Frio/ only in the Nueces, Frio, and Sabinal Rivers on the Sabinal Rivers as well as C. venusta and C. lutrensis (and Edwards Plateau of southwest Texas (Matthews 1987). their hybrids) from the Guadalupe River and tributary This distribution is contained entirely within the ranges of creeks of the Rio Grande. In order to assess which haplo- C. lutrensis and C. venusta. Cyprinella lepida has been types and alleles were unambiguously characteristic of variously described as a subspecies of C. lutrensis (Jordan particular species, additional C. lutrensis from outside the and Evermann 1896; Hubbs 1972) and as a distinct species range of C. venusta in Oklahoma and C. venusta from the 123 Genetica (2011) 139:699–707 701

Fig. 1 Photographs showing phenotypic characteristics of some Cyprinella. a C. lutrensis; b C. venusta; b phenotypic intermediates of C. lutrensis and C. venusta; d C. lepida from the Frio River; e C. lepida from the Nueces River. Photos: A, N. Franssen; B, W. Matthews; C-E, REB

Suwanee River of Georgia were included. Cyprinella for- DNAs were isolated with the DNeasy Kit (Qiagen, Inc.). mosa and C. bocagrande appear to be closely related to We sequenced mitochondrial NADH dehydrogenase sub- C. lutrensis and C. lepida, so several of these individuals unit 2 (ND2) and Cytochrome b (Cyt b) genes and the were also included for phylogenetic perspective. Cypri- intron 1 of the nuclear Hox c6a gene. PCR primers for ND2 nella spiloptera from the Mississippi River basin was used were from Broughton and Gold (2000) and for Cyt b were as the outgroup. from Schmidt et al. (1998). We designed species-specific Most specimens were readily identifiable to species in primers in several cases where apparent nucleotide sub- the field based on diagnostic morphological and coloration stitutions in the priming site inhibited primer binding characters. However, in the Guadalupe River and Syca- (available from R.E.B. on request). Primers for the Hoxc6a more Creek individuals of mixed ancestry between gene were designed for this study using the annotated C. lutrensis and C. venusta were observed. In comparison, genome of Danio rerio on the Ensemble Database. These C. venusta exhibits a long shallow body (typically 9–10 cm were located in exons flanking the single 610 bp intron, standard length) that is mostly silver in coloration with a yielding an amplicon of approximately 800 bp (Hoxc6a- large black spot at the base of the caudal fin in both sexes. int-F 50: gaaccaggtgaagacatttgc 30, Hoxc6a-int-R: 50 cat- Cyprinella lutrensis has a shorter deeper body (typically tggcgatytcgatgcgtc 30). Templates for sequencing were 6–7 cm standard length), red colored fins (except the dorsal amplified in 50 ll reactions containing: 19 reaction buffer

fin), and distinctive blue and red colored post-opercular (10 mM Tris–HCl (pH 8.5), 50 mM KCl, 1.5 mM MgCl2), bars in breeding males. The black caudal spot is absent in 200 lM each dNTP, 0.5 lM each primer, 2.5 U Taq, C. lutrensis. A range of individuals with different combi- approx. 100 ng template DNA, and ultrapure water. Ther- nations of these characters were collected. Some appeared mal cycling conditions were as follows: initial denaturation to be pure of one species or the other, while many inter- at 94° for 1 min., followed by 35 cycles of denaturation at grades appeared to have more or less characteristics of a 94° for 10 s., annealing at 48–55° for 15 s., and polymer- particular species. Some individuals exhibited what ization at 72° for 90 s. Amplification products were puri- appeared to be equal parts of each species as might be fied for sequencing by filtration with Multiscreen PCR96 expected of F1 hybrids. We designated such individuals filter plates (Millipore Inc.). of mixed ancestry as more C. lutrensis-like (lut/ven), more DNA sequencing reactions contained the following in a C. venusta-like (ven/lut), or F1-like (lut=ven). total volume of 10 ll: 1 ll BigDye version 3.1 reaction 123 702 Genetica (2011) 139:699–707

Table 1 Locality information Sample site Species collected Nuclear Mito for specimens used in this study Terlingua Creek, Brewster Co., TX C. lutrensis 05 Tornillo Creek, Brewster Co., TX C. lutrensis 03 Elm Creek, Maverick Co., TX C. lutrensis, C. venusta 14 Sycamore Creek, Cooke Co., TX C. lutrensis, C. venusta, hybrids 5 4 Pinto Creek, Kinney Co., TX C. lutrensis 04 Nueces River, Real/Edwards Co., TX C. lepida 36 Frio River, Real Co., TX C. lepida 43 Sabinal River, Bandera Co., TX C. lepida 22 N Fork Guadalupe River, Kerr Co., TX C. lutrensis, C. venusta, hybrids 4 6 Concho River, Tom Green Co., TX C. lutrensis 25 Colorado River, Concho Co., TX C. lutrensis 01 Croton Creek, Stonewall Co., TX C. lutrensis 11 S Fork Wichita River, Knox Co., TX C. lutrensis 12 N Fork Red River, Greer/Kiowa Co., OK C. lutrensis 10 Stinking Creek, Jackson Co., OK C. lutrensis 02 Washita River, Johnston Co., OK C. lutrensis 03 Little River, Cleveland Co., OK C. lutrensis 10 Two right-hand columns indicate the number of N Canadian River, Woodward Co., OK C. lutrensis 12 specimens sequenced for Morris Creek, Le Flore Co., OK C. lutrensis 03 nuclear and mitochondrial Arkansas River, Muskogee Co., OK C. lutrensis 11 genes, respectively

Fig. 2 Map of major river drainages of Oklahoma and Texas, USA, showing collection localities (with star symbol) for this study. Star fill color indicates species collected: blue = C. lutrensis, black = C. venusta and C. lutrensis, and yellow = C. lepida. State names are underlined; italicized river names are for reference

123 Genetica (2011) 139:699–707 703

Fig. 3 Maximum likelihood tree for the nuclear Hoxc6a intron. Numbers on nodes indicate maximum likelihood bootstrap and Bayesian posterior probabilities, respectively. Individuals are labeled with a species designation based on phenotype followed by the collection locality (from Table 1) and a specimen number. Fishes of mixed phenotype from the Guadalupe River and Sycamore Creek are given the designation of ven/lut if they had more phenotypic characteristics typical of C. venusta, lut/ven if they were phenotypically more like C. lutrensis, or lut=ven if they exhibited characteristics of both species in approximately equal proportions. Bold face indicates individuals with a nuclear genotype different from the species they most resembled phenotypically

mix (Applied Biosystems Inc.), 1.5 ll59 sequencing 2003) using default settings for 2 million generations. For dilution buffer (400 mM Tris–HCl pH 9.0, 10 mM MgCl2), the mitochondrial genes, three data partitions correspond- 0.32 llof10lM primer stock, 3 ll (approx. 50 ng) tem- ing to the three codon positions were used for both maxi- plate DNA, and ultrapure water. Cycling conditions were mum likelihood and Bayesian analyses; the nuclear intron as follows: initial denaturation at 96° for 30 s., followed by was not partitioned. Separate and combined analyses of 40 cycles of 96° for 10 s., 55° for 15 s., and 60° for 4 min. Cytb and ND2 were performed to assess congruence Sequence reaction products were purified via precipitation between mitochondrial genes. with 95% ethanol and 3 M sodium acetate, washed with 70% ethanol, and dried. Electrophoresis and base-calling were performed with an Applied Biosystems 3130xl Genetic Analyzer. Sequences were assembled and aligned Results with Geneious Pro 5.0 (Biomatters Ltd.). Phylogenetic analysis of DNA sequences utilized max- Aligned data matrices contained 1,047 nucleotides (420 imum likelihood and Bayesian methods. Models of DNA phylogenetically informative) for ND2, 1137 (322 phylo- sequence evolution were selected with Modeltest 3.06 genetically informative) for Cyt b, and 800 (53 phyloge- (Posada and Crandall 1998). Maximum likelihood was netically informative) for Hoxc6a intron 1. No heterozygous conducted with the sequential version of RAxML-VI-HPC sites were observed in the intron sequences. The best models ver. 7.0 (Stamatakis 2006). We employed the GTRMIX indicated by ModelTest using both likelihood ratio tests and model wherein sites are partitioned into 25 discrete rate AIC were GTR ? G ? I for each mitochondrial gene and categories for tree searching and topological comparisons, HKY ? G for the nuclear intron. Results of separate phy- and then a discrete gamma distribution is utilized to esti- logenetic analyses of ND2 and Cyt b were identical except mate final tree likelihoods. The ‘‘-a’’ option was used in for a few differences among individuals within the major RAxML which performs non-parametric bootstrapping groups (not shown) so we combined these genes for all (1,000 pseudoreplicates), followed by a search for the subsequent analyses. maximum likelihood tree. Bayesian analyses were per- Results of phylogenetic analyses of the Hoxc6a intron formed with MrBayes 3.1 (Ronquist and Huelsenbeck are shown in Fig. 3. Maximum likelihood and Bayesian

123 704 Genetica (2011) 139:699–707 analyses recovered identical topologies. Several well-sup- Discussion ported groups corresponding to recognized species were recovered although the limited variation in these sequences We examined genealogical patterns focusing on three did not resolve relationships of individuals within species species of Cyprinella with ranges that overlap in the groups. Notably, all individuals of putative C. lepida from American Southwest. Two of these species, C. lutrensis the Nueces and Frio drainages formed a monophyletic and C. venusta, have been known to hybridize in some group which is consistent with the traditional areas of contact and another pair, C. lutrensis and C. lep- recognizing C. lepida as a single taxon. Monophyletic ida, has been suspected of possible hybridization but pre- groups of C. formosa, C. lepida, and C. lutrensis were viously available data have been inconclusive. Our well defined but relationships among these three species approach was to use a phylogenetic framework to compare were not resolved. While not strictly monophyletic, the patterns of possible interbreeding between different species C. venusta-like alleles appear to be closely related to the pairs that are within close geographic proximity. allopatric specimen from the Suwanee River in Georgia. Interbreeding between C. lutrensis and C. venusta is Individuals recovered in the C. lutrensis clade were ini- characterized by the occurrence of individuals of mixed tially identified as pure C. lutrensis or as having more ancestry in several river drainages, indicating multiple C. lutrensis-like characters, in addition to one individual areas of contact and hybridization. In these cases individ- of intermediate phenotype. However, three individuals uals of both parental species are present along with indi- (one from the Guadalupe R. and two from Sycamore Cr.) viduals with intermediate phenotypes (apparent F1 s or exhibiting C. lutrensis-like morphology possessed C. ve- recent backcrosses). The presence of individuals with nusta Hox alleles. Note that while not all the individuals C. lutrensis nuclear alleles and C. venusta mtDNA haplo- are the same as for analysis of the nuclear versus mito- types as well as individuals with C. venusta nuclear alleles chondrial genes, the same set of major drainages are rep- and C. lutrensis mtDNA haplotypes confirms the existence resented and several of the same specimens were used for of hybrid individuals and that hybridization is reciprocal populations suspected of hybridation. between these species. Bidirectional interbreeding appears Phylogenetic analysis of mitochondrial genes also yiel- to produce fertile offspring of both sexes consistent with ded several well-supported groups, however these tended to the experimental results of Hubbs and Strawn (1956). correspond to major river drainages rather than taxonomic Heterospecific mtDNA haplotypes found in both of these designations (Fig. 4). Major clades included C. venusta, species are diverse and broadly representative of the pool C. formosa, C. lepida from the Frio and Sabinal Rivers, of parental haplotypes found in the same geographic region C. lutrensis from the Red River, Arkansas River, and south suggesting that hybridization is not limited to a few rare Texas region, and C. lepida from the Nueces River. There is occurrences. Interspecific genotypic combinations are a large clade that is essentially a four-way polytomy found within a full range of phenotypic intermediates (from including a C. formosa-C. lepida Frio/Sabinal clade, two very lutrensis-like to very venusta-like) suggesting that populations of C. lutrensis from the Arkansas River and Red hybridization is ongoing and that backcrossing may be River drainages and a single individual from the Brazos extensive. Analyses with additional nuclear loci may fur- River of Texas. The C. formosa clade is monophyletic ther characterize the filial identities of individuals in these and includes an individual from the closely related species hybrid zones but there would appear to be separate hybrid C. bocagrande. The relationship between at least some swarms where the species come into contact in these two C. lepida and the formosa group is not unexpected based on south Texas rivers. Nonetheless, there appears to be little previous studies (Broughton and Gold 2000; Schonhuth and introgression outside of the immediate areas of contact. Mayden 2010). The northern populations of C. lutrensis are This situation appears to be somewhat typical of hybrid distinct from those from more southern drainages in Texas zones in other taxa. which flow into the Gulf of Mexico. The south Texas group The genealogical patterns observed for C. lutrensis and forms a clade with the monophyletic C. lepida Nueces C. lepida suggest a very different scenario. The nuclear population embedded within it. data suggest that C. lepida represents a single monophy- A monophyletic C. venusta clade contains one individ- letic species that has diverged to an extent similar to other ual with a C. lutrensis phenotype and one with an inter- recognized species. It is distinct from but related to mediate phenotype, whereas the C. lutrensis South Texas C. lutrensis and C. formosa, as expected based on previous group contains two individuals that were phenotypically phylogenetic studies (Mayden 1989; Broughton and Gold C. venusta-like. Taken with the nuclear result, this suggests 2000; Schonhuth and Mayden 2010). The contrasting that both the Guadalupe River and Sycamore Creek harbor mitochondrial data suggest that the Frio/Sabinal River individuals of mixed ancestry between C. lutrensis and populations may represent original mtDNA lineages of C. venusta. C. lepida, because their relationship with C. formosa and 123 Genetica (2011) 139:699–707 705

Fig. 4 Maximum likelihood tree for mitochondrial ND2 and Cyt b genes. Numbers on nodes indicate maximum likelihood bootstrap and Bayesian posterior probabilities, respectively. Individuals are labeled with a species designation based on phenotype followed by the collection locality (from Table 1) and a specimen number. Fishes of mixed phenotype from the Guadalupe River and Sycamore Creek are given the designation of ven/lut if they had more phenotypic characteristics typical of C. venusta, lut/ven if they were phenotypically more like C. lutrensis, or lut=ven if they exhibited characteristics of both species in approximately equal proportions. Bold face indicates individuals with a mitochondrial haplotype different from their species phenotypic designation

C. lutrensis haplotypes is comparable to that of the nuclear absence of ongoing hybridization where we would expect gene. Alternatively, haplotypes of C. lepida from to observe haplotypes that are not each other’s closest the Nueces River are highly similar to those of the South relatives. The exclusive group of haplotypes combined Texas C. lutrensis population, suggesting that C. lutrensis with a nucleotide diversity comparable to that of the haplotypes have completely replaced the original C. lepida Frio River population, suggests that introgression from haplotypes in this population. Interactions between C. lutrensis was a historic event and that a few introgressed C. lutrensis and C. lepida would only seem to have haplotypes have been diverging in the absence of gene flow occurred at a single locality, the Nueces River, and there is from C. lutrensis since that event. It is possible that other no evidence for hybridization ever occurring in the nearby unsampled genes have introgressed from C. lutrensis into Frio River. Cyprinella lutrensis in not known from the the Nueces population and/or that the Nueces and Frio upper Nueces and individuals with intermediate pheno- populations have diverged from one another. The two types have not been observed. In fact, no hybrid zone populations do appear to exhibit differences in coloration between these two species is known from any locality. The with individuals from the Nueces exhibiting a more gold genotypic combination of C. lepida nuclear alleles with body coloration (Fig. 1d, e). The present data are not suf- C. lutrensis mtDNA haplotypes appears to be fixed in the ficient to address whether there some incompatibility Nueces River. Larger sample sizes will add greater confi- between the Frio population and C. lutrensis or whether the dence to this conclusion, but independent samples from the lack of historical hybridization (or introgression) in the Nueces River by Richardson and Gold (1995) and Schon- Frio is simply due to stochastic factors. It may be that the huth and Mayden (2010) have in no case recovered indi- two populations of C. lepida have diverged and represent viduals that are not consistent with a population fixed for different species but the relatively conserved Hox intron heterospecific mtDNA. does not reflect this divergence. Heterospecific mtDNAs that are fixed, monophyletic, Ultimately, the taxonomic status of C. lepida popula- and exclusive to C. lepida of the Nueces suggests the tions remain controversial. Matthews (1987) noted

123 706 Genetica (2011) 139:699–707 coloration differences between the Nueces and Frio popu- Undergraduate Research Opportunity awards from the University of lations and extensive divergence of mtDNA led Richardson Oklahoma Honors College (to K.C.V. and L.L.R.) and award DEB- 0732988 from the National Science Foundation (to R.E.B.). and Gold (1995) to suggest that the Nueces population was a new species. In an extensive phylogenetic study of References Cyprinella, Schonhuth and Mayden (2010) recovered two distinctive populations based on Cyt b sequences that they Aboim MA, Mavarez J, Bernatchez L, Coelho MM (2010) Introgres- referred to as C. lepida (from the Frio) and C. sp. ‘cf. sive hybridization between two Iberian endemic cyprinid fish: a lepida’ (from the Nueces). However, analysis of the comparison between two independent hybrid zones. J Evol Biol nuclear Rag1 gene for two individuals from the Frio and 23:817–828 one from the Nueces recovered all three as monophyletic Alves MJ, Coelho MM, Collares-Pereira MJ (2001) Evolution in action through hybridization and polyploidy in an Iberian with strong support. Schonhuth and Mayden (2010) sug- freshwater fish: a genetic review. Genetica 111:375–385 gested introgression of mtDNA in the Nueces population Blum MJ, Walters DM, Burkhead NM, Freeman BJ, Porter BA but its taxonomic status remains undefined. Resolution of (2010) Reproductive isolation and the expansion of an invasive this question will require characterization of additional hybrid swarm. Biol Invasions 12:2825–2836 Bolnick DI, Near TJ (2005) Tempo of post-zygotic reproductive nuclear genes in the context of whether this population isolation in sunfishes (Teleostei: Centrarchidae). Evolution represents a new species, possibly of hybrid origin. The 59:1754–1767 central issue might be whether introgression of mtDNA Broughton RE and Gold JR (2000) Phylogenetic relationships in the alone is sufficient to describe a new species. North American cyprinid genus Cyprinella (: Cyprinidae) based on sequences of the mitochondrial ND2 and The frequency of hybridization among freshwater fishes ND4L genes. Copeia 1–10 may be due in part to functional divergence of populations Campton DE (1987) Natural hybridization and introgression in fishes: in relative isolation in different streams or rivers. The Methods of detection and genetic interpretations. In: Ryman N, ability to interbreed appears to be relatively conserved Utter F (eds) Populations genetics and fishery management. University of Washington Press, Seattle, pp 161–192 among freshwater taxa suggesting that traits that may affect DeMarais BD, Dowling TE, Douglas M, Minckley WL, Marsh P post-zygotic reproductive isolation evolve more slowly and (1992) Origin of Gila seminuda (Teleostei: Cyprinidae) through are more or less uncoupled from functionally adaptive introgressive hybridization: Implications for evolution and traits. The habit of many cyprinid species to spawn in conservation. Proc Nat Acad Sci USA 89:2747–2751 Dowling TE, Childs MR (1992) Impact of hybridization on a groups over crevices or the nests of other species reflects threatened trout of the Southwestern United States. Conserv Biol weak prezygotic barriers in many groups. It may be that 6:355–364 physical separation is sufficient to keep most mid-reach Dowling TE, Secor CL (1997) The role of hybridization in the and headwater species isolated and there has been little evolutionary diversification of . Ann Rev Ecol Syst 28:593–619 adaptive value to reproductive characters that evolve rap- Dowling TE, Smith GR, Brown WM (1989) Reproductive isolation idly. This may explain the prevalence of hybridization between cornutus and Notropis chrysocephalus (family where species come into contact. Human activities and Cyprinidae): Comparison of morphology, allozymes, and mito- habitat alterations may in fact be increasing contact chondrial DNA. Evolution 43:620–634 Dowling TE, Broughton RE, DeMarais BD (1997) Significant role for between species that might otherwise remain separated and historical effects in the evolution of reproductive isolation: may contribute to the frequency of hybridization. We have evidence from patterns of introgression between the cyprinid shown that hybridization between different pairs of Cyp- fishes, Luxilus cornutus and Luxilus chrysocephalus. Evolution rinella species may have different outcomes even in a 51:1574–1583 Echelle AA, Connor PJ (1989) Rapid, geographically extensive genetic narrow geographic area. Evidence for historical hybrid- introgression after secondary contact between two pupfish species ization suggests that it must have occurred at some fre- (Cyprinodon, Cyprinodontidae). Evolution 43:717–727 quency prior to human influence. Further genetic Gerber AS, Tibbets CA, Dowling TE (2001) The role of introgressive characterization of individuals within current and historical hybridization in the evolution of the Gila robusta complex (Teleostei: Cyprinidae). Evolution 55:2028–2039 hybrid zones as well as the ecological factors that charac- Harrison RG (1990) Hybrid zones: windows on evolutionary process. terize these contact zones will provide new insights on Oxf Surv Evol Biol 7:69–128 speciation and perhaps further illuminate the factors that Hubbs CL (1955) Hybridization between fish species in nature. Syst make hybridization so common in freshwater fishes. Zool 4:1–20 Hubbs C (1972) A checklist of Texas freshwater fishes. Tex Parks Wildl Dept Tech Ser 11:1–11 Acknowledgments R.E.B. wishes to thank Richard G. Harrison for Hubbs C, Strawn K (1956) Interfertility between two sympatric fishes, excellent advice and inspiration. We thank E. Marsh-Matthews, W. Notropis lutrensis and Notropis venustus. Evolution 10:341–344 Matthews and P. Reneau for assistance collecting or providing Jordan DS, Evermann BW (1896) The fishes of North and Middle specimens. All specimens were collected under the relevant permits America. Bull US Nat Hist Mus 47:1–1240 granted by the states of Texas and Oklahoma. We thank E. Marsh- Keck BP, Near TJ (2010) Geographic and temporal aspects of Matthews, W. Matthews, J. Gold and L. Richardson for insightful mitochondrial replacement in Nothonotus darters (Teleostei: discussions about southwestern Cyprinella. The work benefited from Percidae: Etheostomatinae). Evolution 64:1410–1428

123 Genetica (2011) 139:699–707 707

Lytle GL (1972) Cyprinid fishes of the subgenus Cyprinella of Schonhuth S, Mayden RL (2010) Phylogenetic relationships in the Notropis from southeast Texas, USA., and northeast Mexico, genus Cyprinella (Actinopterygii: Cyprinidae) based on mito- M.S. Thesis, Arizona State University, Tempe, AZ chondrial and nuclear gene sequences. Mol Phylogenet Evol Matthews WJ (1987) Geographic variation in Cyprinella lutrensis 55:77–98 (Pisces: Cyprinidae) in the United States, with notes on Scribner KT, Page KS, Bartron ML (2001) Hybridization in Cyprinella lepida. Copeia 1987:616–637 freshwater fishes: a review of case studies and cytonuclear Mayden RL (1989) Phylogenetic studies of North American min- methods of biological inference. Rev fish biol fish 10:293–323 nows, with emphasis on the genus Cyprinella (Teleostei: Smith GR (1992) Introgression in fishes significance for paleontology, ). Univ Kansas Mus Nat Hist Misc Publ 80:1–189 cladistics, and evolutionary rates. Syst Biol 41:41–57 Mayden RL, Burr BM, Page LM, Miller RR (1992) The native Stamatakis A (2006) RAxML-VI-HPC: Maximum likelihood-based freshwater fishes of North America. In: Mayden RL (ed) phylogenetic analyses with thousands of taxa and mixed models. Systematics historical ecology and North American freshwater Bioinformatics 22:2688–2690 fishes. Stanford University Press, Stanford, pp 827–863 Turner BJ, Brett BH, Miller RR (1980) Interspecific hybridization and Posada D, Crandall KA (1998) Modeltest: testing the model of DNA the evolutionary origin of a gynogenetic fish, Poecilia formosa. substitution. Bioinformatics 14:817–818 Evolution 34:917–922 Richardson LR and Gold JR (1995) Evolution of the Cyprinella- Walters DM, Blum MJ, Rashleigh B, Freeman BJ, Porter BA, Lutrensis Species-Complex .2. Systematics and Biogeography of Burkhead NM (2008) invasion and hybridization with the Edwards , Cyprinella-Lepida. Copeia 28–37 in the upper Coosa River, USA. Biol Invas Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phyloge- 10:1229–1242 netic inference under mixed models 19:1572–1574 Schmidt TR, Bielawski JP, Gold JR (1998) Molecular phylogenetics and evolutionof the cytochrome b gene in the cyprinid genus Lythrurus (Actinopterygii: Cypriniformes). Copeia 1998:14–22

123